Automating Spinnaker Deployments w/ Halyard


How are users doing this now? How much of a guide should we write up? What are weak points in current setups?


I’d love to hear about this too. Right now when engaging with customers they consistently want to be able to express the whole deployment as code so deployments can be torn down and reconstructed at will.


Currently we are using Ansible for setting up and destroying our spinnaker on gce it is mixture of hal commands and some shell commands. I have tried compiling bit of it here though it is just specific for GCE for now. Feel free to try it out @morgantep , any kind of feedback and suggestion is welcome.


We’re using Packer to build AMIs to get config and packages onto the box, with a simple upstart to run a hal deploy apply and attach to the appropriate target groups on instance launch. Provisioners are just shell for the moment though we’ll likely move to puppet once we’re templating all the config instead of running hal commands.